As the voice of the consumer, social media provides uninhibited, real-time access to the consumer and their experiences with fintechs, banks, and other financial providers. Never before has such a direct pulse on consumer sentiment existed to help regulators and others quickly identify trends or immediate problems.
Though a powerful new addition to the regulatory toolbox over the last decade, the monitoring of social media and online forums has key limitations. Some areas where it falls short:
- Self-reported by select consumers
- Lends itself to qualitative analysis, rather than quantitative analysis
- Details vary as there is no standard template for consumers
- Tone isn’t easily detected (at least yet) by machine learning
Those who take to social media and forums to speak about their financial experiences are a self-reporting and often self-selecting crew. These customers that raise their voice may have differing levels of education, wealth, financial literacy, etc. as compared to the more general population. Listening to these voices should not preclude regulators from listening to all types of consumer voices and ensuring it has other venues to do so (e.g., town halls, meeting and touring third-party consumer help organizations, focus groups, etc.).
In this vein, analysis of social media data is often better suited for qualitative analysis rather than quantitative ones. For example, there are significantly more social media posts regarding consumer’s experiences with banks as opposed to insurance providers. Taking a quantitative approach might lead one to inaccurately believe that banks are multiples more risky when the reality is that the volume of posts more directly ties with how consumers engage with the company than with the underlying product: consumers interact daily with their banks but may only engage yearly with their insurance provider. Rather, a qualitative approach of understanding the voice and nature of consumer comments within each sector and even at the firm level yields more generalizable insights than would a quantitative one.
Lastly, social media and web data largely come in unstructured formats with varying levels of details provided by consumers. For automated analyses, machine learning techniques are increasingly used; however, these techniques have not yet been perfected. Detecting emotional tone like sarcasm remains a challenge. It can be difficult for algorithms to analyze web content that is in or across multiple languages in a single jurisdiction as well. Consequently, some level of manual review is likely to remain for the foreseeable future.
What this means for fintechs and regulators
The capability to use social media and web scraping to monitor consumer sentiment, consumer experiences and the reputations of financial providers is clearly additive to regulators’ toolboxes.
As a fintech or bank, we should expect regulators to increasingly reference social media and online posts in our engagements. We should be equally prepared to educate regulators and others on its proper usage for purposes of supervision.
To be clear, monitoring social media is not a silver bullet and should be complementary to other means of market intelligence and data collection. Hearing directly from consumers in their own voices - through social media, web scraping, and complaint data - is just one vantage of many into the state of the financial marketplace for regulators to view.